On 4 Aug, Tom Lane wrote:
> [ Snip discussion regarding PQconnectdb()s thread-safety ]
>
> If you want to work on it, be my guest...
I don't have time to think about this today, so I can't comment on how
it should work, but I _am_ currently working in this area - I am
providing non-blocking versions of the connect statements, as discussed
on the interfaces list a couple of weeks ago. In fact, it is pretty
much done, apart from a tidy-up, documentation, and testing. I don't
see any point in two people hammering away at the same code - it will
only make work when we try to merge again - so perhaps I should
implement what ever is decided - I don't mind doing so. However, if I
didn't get it done this weekend it would have to be mid-to-late
September, since I'm going away. Would that be a problem for anyone?
I had noticed that the connect statements weren't thread-safe, but
was neither aware that that was a problem for anyone, nor inclined to
audit the whole of libpq for thread-safety, so I left it alone.
Ewan.